Recording backing vocals is a little different to recording the lead vocals. Rather than recording them forwards, then backwards, then forwards, I simply record them one section at a time – typically four or six takes for each part. I prefer a combination of syncronised harmony vocals (in time and harmony with the lead vocal) and unsynchronised ‘call and response’-type backing vocals (with different timing and rhythm to the lead vocal).
For bigger backing vocals, I’ll take the two best takes for each part, and pan them hardleft and hardright. The natural variance in intonation gives the part a very wide sound without being messy. It also sounds much more natural than using a single take and making widening it using artificial processes (such as delays or pitch shifting). When I want even more voices, I record different harmony parts and apply the same process. Sometimes I’ll go as many as three parts deep. This results in six total harmony tracks – three on each side.
The trick with harmony vocals is to go easy on intonation correction. Whether you use Autotune, Melodyne, GSnap, or something else, find a way to use it extremely subtly. The more in-tune the backing vocals are, the smaller the total effect is. Your job is to balance correctness with size. I find even correcting the vocals 50% has a significant effect – often too much! A lot of the time I’m happy to keep the backing vocals untuned, or tune one side and keep the other untuned. So long as the singer can sing reasonably well, it shouldn’t be too detrimental to the song. If the lead vocal is appropriately in tune, the backing vocals only need to add size and thickness.
On the other hand, if the lead vocal is weak (tuned or not), the backing vocals benefit from being much more in tune. In this situation, the backing vocals serve as a support for the vocal (and should be mixed appropriately).